Research map / Graph learning / GNN architectures
U2B: Scale-unbiased Representation Converter for Graph Classification with Imbalanced and Balanced Scale Distributions
Guanjun Wang, Jianhao Zhang, Jiaming Ma, Sheng Huang, Pengkun Wang, Zhengyang Zhou, Binwu Wang, Yang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add U2B to GIN | GIN · PROTEINS (scale-imbalanced) | Macro-F1: 53.48 -> 83.14 | Improved |
| add U2B to GCN | GCN · PROTEINS (scale-imbalanced) | Macro-F1: 51.29 -> 74.65 | Improved |
| add U2B to GraphSAGE | GraphSAGE · PROTEINS (scale-imbalanced) | Macro-F1: 52.95 -> 78.43 | Improved |
| add U2B to Exphormer | Exphormer · NCI1 (scale-imbalanced) | Macro-F1: 65.16 -> 78.48 | Improved |
| add U2B to G-Mamba | G-Mamba · NCI1 (scale-imbalanced) | Macro-F1: 64.09 -> 82.29 | Improved |
| add node-level converter (ablation) | U2B · PROTEINS (scale-imbalanced) | F1-Score | Improved |
| add graph-level converter (ablation) | U2B · scale-imbalanced datasets | F1-Score | Improved |
| add Sinkhorn-Knopp static constraint (ablation) | U2B · scale-imbalanced datasets | F1-Score | Improved |
| add dynamic constraint in graph-level converter (ablation) | U2B · scale-imbalanced datasets | F1-Score | Improved |
| add consistency alignment loss (ablation) | U2B · scale-imbalanced datasets | F1-Score | Improved |
| increase distilled token numbers Kv and Kg beyond 16-64 (ablation) | U2B · scale-imbalanced datasets | F1-Score | Hurt |
| use U2B instead of GIN on balanced data | GIN · IMDB-MULTI (scale-balanced) | Macro-F1: 35.91 -> 46.46 | Improved |
Closest papers by meaning
- Subgraph Invariant Learning Towards Large-Scale Graph Node Classification (AAAI 2025 · similarity 0.87)
- Class-Imbalanced Graph Learning without Class Rebalancing (ICML 2024 · similarity 0.87)
- IGL-Bench: Establishing the Comprehensive Benchmark for Imbalanced Graph Learning (ICLR 2025 · similarity 0.87)
- Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized Approach (AAAI 2025 · similarity 0.86)
- Cluster-guided Contrastive Class-imbalanced Graph Classification (AAAI 2025 · similarity 0.86)
- Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
